SYnthetic data Generation

The volume, velocity and variety of national security threats are increasing at an alarming rate. Fears of terrorism, cyberattacks, and WMD proliferation permeate all levels of society. Decision makers are placing increased demands on resource-constrained intelligence analysts to anticipate evolving and emerging threats hidden in a deluge of real-time streaming data across a variety of sources. Analysts need big data analytics software solutions that leverage advanced mathematical and computational techniques for modeling adversarial activity to produce high-confidence indications and warnings of threats. 

It is, however, extremely difficult to find meaningful test data for evaluating big data analytics software solutions that detect adversarial efforts to plan and execute terrorist, cyber, and/or WMD attacks. Relevant threat activity continuum data may not be readily available or may be classified, thereby restricting access. Unclassified, representative test data sets containing known ground truth regarding critical threat scenarios are needed to quantitatively evaluate the performance of big data analytics tools. 

IvySys is developing synthetic data generation tools to create realistic, large-scale data sets that provide synthetic threat transactions seamlessly blended within a realistic background environment. Our tools will automatically create threat scenario-based temporal transaction data sets across multiple data channels. The synthetic data generation tools will allow users to vary the threat detection difficulty level during the data set generation process. This ability to produce data sets over a range of threat signal-to-noise ratios enables robust performance testing of threat detection analytics software tools.

Additionally, IvySys is developing big data analytics tools to detect terrorist, cyber and WMD threat activities buried in dense background environments. These tools use graph signal processing and machine learning techniques to enable large-scale graph analytics. The IvySys advanced analytics algorithms use graph alignment and merging methods to integrate multiple data sources, as well as sub-graph detection and sub-graph matching methods. This approach allows users to quickly identify suspicious (or anomalous) activities and events in noisy, complex, environments.

IvySys big data and analytics solutions leverage the following core competencies:

• Synthetic big data generation

• Activity modeling and simulation

• Machine learning

• Automated indicators and warnings

• Big data visualization

• Graph analytics

• Multi-source spatial temporal data correlation and fusion

• Data cleansing